Neural Signatures of Predictive Processing under Central vs Peripheral Noradrenergic Blockade

927 

Abstract Submission 

Ashley Tyrer¹, Micah Allen²

¹Aarhus University, Aarhus, Aarhus C, ²Aarhus University, Lystrup, Denmark

Ashley Tyrer  
Aarhus University
Aarhus, Aarhus C

Micah Allen, Professor, PhD  
Aarhus University
Lystrup, Denmark

While considerable research has been conducted regarding the effects of noradrenergic manipulations on uncertainty and prediction, little is known about the differential effects on cognitive functioning of central vs peripheral noradrenergic blockade, and how such manipulations may impact the processing of uncertainty in contrasting ways – particularly in the context of magnetoencephalography (MEG). Further, little is known about how such manipulations may affect brain-body interactions in response to uncertainty and prediction error.
Here, we employed task-based MEG, pharmacological manipulation and pupillometry in a double-blinded, placebo-controlled, within-subject context to investigate how the processing of uncertainty and prediction may be influenced by central vs peripheral beta-adrenoceptor antagonism in 50 young healthy participants. We aimed to elucidate the neural and embodied signatures of (unexpected) uncertainty, and how these may be altered by changes in central or peripheral noradrenaline. Therefore, we expected participants across both conditions to exhibit slower learning rates and belief updating, reduced internal model flexibility, and stronger weighting of prior beliefs in decision-making, in line with previous analyses.
In addition to examining neural activity in response to uncertainty, we also examined a variety of physiological responses such as cardiac and respiratory signals to probe how brain-body responses to uncertainty are altered in central and peripheral noradrenaline manipulation.

Fifty young healthy participants attended three separate study visits each in this double-blinded, placebo-controlled, within-subject study, where each study visit consisted of a single acute drug administration (40 mg propranolol, 10 mg bisoprolol, or a matched placebo) followed by simultaneous MEG imaging, physiological recordings and pupillometry, during which participants completed a perceptual reversal learning task. This task consisted of a revised version of the common probabilistic associative learning task with the addition of confidence ratings at the end of each trial and was preceded by individual staircasing of visual stimuli, i.e., emotional faces. Further, following the MEG scan participants completed 60 minutes of cognitive tasks on a computer outside the scanner. Participants completed brief testing measures of their interoceptive sensitivity, in both the cardiac and respiratory domain, namely the Heart Rate Discrimination task (HRD) and the Respiratory Resistance Sensitivity Task (RRST).

We found a significantly attenuated neural response to task-related prediction error in the propranolol condition compared with the placebo condition. We also found that greater metacognitive ability was reflected in greater task prediction accuracy and found this to be improved under propranolol, in line with previous work. We also noted a reduced attentional response to stimuli and a blunted response to prediction errors under propranolol but not bisoprolol. This speaks to the centrally mediated nature of saliency detection, and therefore enabled us to better tease apart the neural representations of emotional stimuli in the bisoprolol condition. Additionally, through the use of Active Inference modelling of behavioural responses we generated subject-specific conditional estimates of internal model volatility under each drug condition. Further, under bisoprolol, we found a significantly reduced cardiac response to uncertainty compared with the placebo condition.

Overall, we show here that neural activations in task-relevant brain regions associated with uncertainty and predictive processing, such as the anterior cingulate cortex, are altered in response to belief updating and model volatility, differentially for central vs peripheral noradrenergic manipulation. We also demonstrate subject-specific neural and behavioural profiles of internal model volatility and predictive processing in response to uncertainty.

Executive Function, Cognitive Control and Decision Making ¹

Bayesian Modeling

EEG/MEG Modeling and Analysis

MEG ²

Pharmacology and Neurotransmission

Cognition

Computational Neuroscience

MEG

Neurotransmitter

Noradrenaline